Drug impurity profiling strategies

A general scheme is set up for the estimation of the impurity profile of bulk drug substances by the complex use of chromatographic, spectroscopic and hyphenated techniques. Several examples are presented as illustrations to the scheme from the authors' laboratory involving the use of chromatographic methods such as thin-layer-(TLC), gas-(GC), analytical and preparative high-performance liquid chromatography (HPLC), spectroscopic methods such as mass spectrometry (MS) and NMR spectroscopy as well as hyphenated techniques (HPLC/diode-array UV, GC/MS and HPLC/MS). In addition to summarizing earlier work, new examples are also presented: identification of an impurity (propyl 4-[diethylcarbamoyl(methoxy)]-3-methoxy phenylglyoxylate, II) in propanidid (I) and two unsaturated impurities in allylstrenol (VII) by GC/MS and HPLC/diode-array UV as well as estimation of the impurity profile of mazipredone (III) by HPLC/MS and HPLC/diode-array UV.     

Drug profiling using planar microelectrode arrays

Microelectrode arrays (MEAs) with evenly distributed multiple sensor spots have been designed for specific applications. Using the MEAs, we determined the relative profiles of potassium channel openers (KCOs) on cultured embryonic Sprague-Dawley rat cardiac myocytes. KCO, pinacidil (PIN), cromakalim (CROM), SDZ PCO400 (SDZ), or its vehicle, was added to the myocytes cumulatively. The action potential signal shapes in the presence of PIN and SDZ show that the changes in voltage over time and the magnitudes of the associated voltage change were reduced concentration-dependently. CROM affected sodium influx more than PIN and SDZ. The comparisons of changes in the rate of beating and propagation speed in the presence of KCOs were made using their corresponding pD(2) values (the negative log of EC(50)). All KCOs caused concentration-dependent reductions in the rate of beating and propagation speed, with SDZ being the most potent. In addition to the signal shapes, rate of beating, and propagation speed, the origin of excitation and the excitation pattern inside the culture can be also extracted. The results show that the present system can differentiate the effects of different KCOs on myocytes. It might be possible to utilise the MEA as a means to classify drug action based upon a combined interpretation of the three different datasets gained from the extracellular recordings. The combination of these observations might be used as 'drug signatures' when profiling drugs in the future.     

Ligand binding affinity prediction by linear interaction energy methods

A recent method for estimating ligand binding affinities is extended. This method employs averages of interaction potential energy terms from molecular dynamics simulations or other thermal conformational sampling techniques. Incorporation of systematic deviations from electrostatic linear response, derived from free energy perturbation studies, into the absolute binding free energy expression significantly enhances the accuracy of the approach. This type of method may be useful for computational prediction of ligand binding strengths, e.g., in drug design applications.     

Drug impurity profiling by capillary electrophoresis/mass spectrometry using various ionization techniques

Capillary electrophoresis/mass spectrometry (CE/MS) is predominantly carried out using electrospray ionization (ESI). Recently, atmospheric pressure chemical ionization (APCI) and atmospheric pressure photoionization (APPI) have become available for CE/MS. With the VUV lamp turned off, the APPI source may also be used for CE/MS by thermospray ionization (TSI). In the present study the suitability of ESI, APCI, APPI and TSI for drug impurity profiling by CE/MS in the positive ion mode is evaluated. The drugs carbachol, lidocaine and proguanil and their potential impurities were used as test compounds, representing different molecular polarities. A background electrolyte of 100 mM acetic acid (pH 4.5) provided baseline separation of nearly all impurities from the respective drugs. APPI yielded both even‐ and odd‐electron ions, whereas the other ionization techniques produced even‐electron ions only. In‐source fragmentation was more pronounced with APCI and APPI than with ESI and TSI, which was most obvious for proguanil and its impurities. In general, ESI and TSI appeared the most efficient ionization techniques for impurities that are charged in solution achieving detection limits of 100 ng/mL (full‐scan mode). APPI and APCI showed a lower efficiency, but allowed ionization of low and high polarity analytes, although quaternary ammonium compounds (e.g. carbachol) could not be detected. Largely neutral compounds, such as the lidocaine impurity 2,6‐dimethylaniline, could not be detected by TSI, and yielded similar detection limits (500 ng/mL) for ESI, APPI and APCI. In many cases, impurity detection at the 0.1% (w/w) level was possible when 1 mg/mL of parent drug was injected with at least one of the CE/MS systems. Overall, the tested CE/MS systems provide complementary information as illustrated by the detection and identification of an unknown impurity in carbachol. Copyright © 2009 John Wiley & Sons, Ltd.     

Kernel-based data fusion improves the drug-protein interaction prediction

Proteins are involved in almost every action of every organism by interacting with other small molecules including drugs. Computationally predicting the drug-protein interactions is particularly important in speeding up the process of developing novel drugs. To borrow the information from existing drug-protein interactions, we need to define the similarity among proteins and the similarity among drugs. Usually these similarities are defined based on one single data source and many methods have been proposed. However, the availability of many genomic and chemogenomic data sources allows us to integrate these useful data sources to improve the predictions. Thus a great challenge is how to integrate these heterogeneous data sources. Here, we propose a kernel-based method to predict drug-protein interactions by integrating multiple types of data. Specially, we collect drug pharmacological and therapeutic effects, drug chemical structures, and protein genomic information to characterize the drug-target interactions, then integrate them by a kernel function within a support vector machine (SVM)-based predictor. With this data fusion technology, we establish the drug-protein interactions from a collections of data sources. Our new method is validated on four classes of drug target proteins, including enzymes, ion channels (ICs), G-protein couple receptors (GPCRs), and nuclear receptors (NRs). We find that every single data source is predictive and integration of different data sources allows the improvement of accuracy, i.e., data integration can uncover more experimentally observed drug-target interactions upon the same levels of false positive rate than single data source based methods. The functional annotation analysis indicates that our new predictions are worthy of future experimental validation. In conclusion, our new method can efficiently integrate diverse data sources, and will promote the further research in drug discovery.     

Molecule-ion interaction and its effect on coordination interaction

The present work revealed the presence of the molecule-ion interaction between ethylenediaminetetraacetic acid disodium salt (Na(2)H(2)EDTA) and β-cyclodextrin (CD) on the basis of observable changes in crystal patterns and thermal behaviors before and after interaction. Results from electric conductivity measurements confirmed this presence and showed that the extent of the molecule-ion interaction was associated with the concentration of β-CD. More importantly, the molecule-ion interaction led to a decreased coordination interaction of Na(2)H(2)EDTA and copper chloride, and this decrease exhibited a concentration dependence of β-CD. Similar phenomena were also observed in the case of several analogs of Na(2)H(2)EDTA by UV-vis spectroscopy. A possible explanation was proposed on the basis of the hypothesis that there was a competitive relationship between the molecule-ion interaction and the coordination interaction. Further, nuclear magnetic resonance measurements provided important information on the difference in interaction modes of β-CD with H(2)EDTA(2-) and [Cu(EDTA)](2-). We are of the opinion that the results would provide a significant bridge between coordination chemistry and supramolecular chemistry and help us further understand factors related to different interactions in multicomponent systems.     

Proteomic dissection of biological pathways/processes through profiling protein-protein interaction networks

Cellular functions, either under the normal or pathological conditions or under different stresses, are the results of the coordinated action of multiple proteins interacting in macromolecular complexes or assemblies. The precise determination of the specific composition of protein complexes, especially using scalable and high-throughput methods, represents a systematic approach toward revealing particular cellular biological functions. In this regard, the direct profiling protein-protein interactions (PPIs...     

Metabolite profiling of sucrose effect on the metabolism of Melissa officinalis by gas chromatography-mass spectrometry

The effect of sugar on plant metabolism, which is known to be similar to hormone-like signaling, was metabolomically studied using Melissa officinalis (lemon balm). The metabolite profiles of M. officinalis treated with sucrose were analyzed by gas chromatography-mass spectrometry (GC-MS) and principal component analysis (PCA). A total of 64 metabolites from various chemical classes including alcohols, amines, amino acids, fatty acids, inorganic acids, organic acids, phosphates, and sugars were identified by GC-MS. Three groups treated with different sucrose concentrations were clearly separated by PCA of their metabolite profiles, indicating changes in the levels of many metabolites depending on the sucrose concentration. Metabolite profiling revealed that treatment with a higher sucrose level caused an increase in the levels of metabolites such as sugars, sugar alcohols, and sugar phosphates, which are related to the glycolytic pathway of M. officinalis. Furthermore, proline and succinic acid, which are associated with the proline-linked pentose phosphate pathway, the shikimic acid pathway, and the biosynthesis of phenylpropanoids, also increased with increasing sucrose concentration. Therefore, these metabolic changes induced by sucrose ultimately led to the increased production of flavonoids such as caffeic acid via the biosynthetic pathway of phenylpropanoids. This study demonstrated that the abundance changes in some primary and secondary metabolites were somewhat interlocked with each other in response to sucrose.     

Characterization of interaction kinetics between chiral solutes and human serum albumin by using high-performance affinity chromatography and peak profiling

Peak profiling and high-performance columns containing immobilized human serum albumin (HSA) were used to study the interaction kinetics of chiral solutes with this protein. This approach was tested using the phenytoin metabolites 5-(3-hydroxyphenyl)-5-phenylhydantoin (m-HPPH) and 5-(4-hydroxyphenyl)-5-phenylhydantoin (p-HPPH) as model analytes. HSA columns provided some resolution of the enantiomers for each phenytoin metabolite, which made it possible to simultaneously conduct kinetic studies on each chiral form. The dissociation rate constants for these interactions were determined by using both the single flow rate and multiple flow rate peak profiling methods. Corrections for non-specific interactions with the support were also considered. The final estimates obtained at pH 7.4 and 37°C for the dissociation rate constants of these interactions were 8.2-9.6 s(-1) for the two enantiomers of m-HPPH and 3.2-4.1 s(-1) for the enantiomers of p-HPPH. These rate constants agreed with previous values that have been reported for other drugs and solutes that have similar affinities and binding regions on HSA. The approach used in this report was not limited to phenytoin metabolites or HSA but could be applied to a variety of other chiral solutes and proteins. This method could also be adopted for use in the rapid screening of drug-protein interactions.     

GanDTI: A multi-task neural network for drug-target interaction prediction

Drug discovery processes require drug-target interaction (DTI) prediction by virtual screenings with high accuracy. Compared with traditional methods, the deep learning method requires less time and domain expertise, while achieving higher accuracy. However, there is still room for improvement for higher performance with simplified structures. Meanwhile, this field is calling for multi-task models to solve different tasks. Here we report the GanDTI, an end-to-end deep learning model for both interaction classification and binding affinity prediction tasks. This model employs the compound graph and protein sequence data. It only consists of a graph neural network, an attention module and a multiple-layer perceptron, yet outperforms the state-of-the art methods to predict binding affinity and interaction classification on the DUD-E, human, and bindingDB benchmark datasets. This demonstrates our refined model is highly effective and efficient for DTI prediction and provides a new strategy for performance improvement.     

Glycoprotein profiling by electrospray mass spectrometry

This work compares several different methods of site-specific analysis of glycoproteins using electrospray mass spectrometry. The glycoprotein, oLHalpha (ovine luteinizing hormone, alpha-subunit) was chosen as an appropriate example protein for these studies because of its biological relevance and extreme microheterogeneity. More than 20 unique glycoforms were detected for this glycoprotein at the Asn(56) site of oLHalpha. The carbohydrates present at this site affect receptor binding affinity, so understanding the great variety in the composition of these carbohydrates is important in studying ligand binding interactions. MS data was acquired on a quadrupole ion trap, a triple quadrupole, and a quadrupole time of flight mass spectrometer, and carbohydrate composition at the Asn(56) site of oLHalpha was determined using these instruments. Additionally, neutral loss and precursor ion scanning modes were also used to identify the glycoforms present, and these techniques were compared to the standard MS data. Of the three instruments compared in the study, the qTOF mass spectrometer achieved the lowest sample consumption, but all three instruments were useful in profiling the glycopeptide composition.     

Ancient DNA profiling by megaplex amplications

Simultaneous amplification of nine human short tandem repeat (STR) DNA sequences and the amelogenin locus allows reducing to an absolute minimum the amount of sample material that is necessary for genetic identification or kinship analysis. Valuable remains can be studied this way without any visible damage, as is demonstrated by typing the DNA of a tooth root from the Saxon warrior Widukind, who died about 1200 years ago. The broad applicability of the megaplex approach is shown by typing bone and teeth specimens ranging from a few months to 3000 years of age employing AmpFlSTR Profiler Plus. Additionally, megaplex STR typing is the method of choice for proving the authenticity of molecular results derived from ancient degraded DNA.     

Learning size-adaptive molecular substructures for explainable drug–drug interaction prediction by substructure-aware graph neural network

Drug–drug interactions (DDIs) can trigger unexpected pharmacological effects on the body, and the causal mechanisms are often unknown. Graph neural networks (GNNs) have been developed to better understand DDIs. However, identifying key substructures that contribute most to the DDI prediction is a challenge for GNNs. In this study, we presented a substructure-aware graph neural network, a message passing neural network equipped with a novel substructure attention mechanism and a substructure–substructure interaction module (SSIM) for DDI prediction (SA-DDI). Specifically, the substructure attention was designed to capture size- and shape-adaptive substructures based on the chemical intuition that the sizes and shapes are often irregular for functional groups in molecules. DDIs are fundamentally caused by chemical substructure interactions. Thus, the SSIM was used to model the substructure–substructure interactions by highlighting important substructures while de-emphasizing the minor ones for DDI prediction. We evaluated our approach in two real-world datasets and compared the proposed method with the state-of-the-art DDI prediction models. The SA-DDI surpassed other approaches on the two datasets. Moreover, the visual interpretation results showed that the SA-DDI was sensitive to the structure information of drugs and was able to detect the key substructures for DDIs. These advantages demonstrated that the proposed method improved the generalization and interpretation capability of DDI prediction modeling.

SA-DDI is designed to learn size-adaptive molecular substructures for drug–drug interaction prediction and can provide explanations that are consistent with pharmacologists.     

Backscattering effect in quantitative AES sputter depth profiling of multilayers

AES sputter depth profiles of multilayers with constituents of very different backscattering factors show characteristic distortions in the shape of the intensity–depth profiles. These distortions are quantified by introducing an extension of the local effective backscattering factor concept developed in an earlier paper in the mixing‐roughness‐information depth (MRI) model for profile quantification. The extension is based on a linear superposition of two newly defined parameters, the effective backscattering factors for each interface that are diminished with distance from the respective interface by another characteristic parameter, the mean effective backscattering decay length. As shown for a Ni/C multilayer structure of six alternating layers of Ni (38 nm) and C (25 nm) on a Si substrate, AES intensity depth profiles calculated with the presented modification of the MRI model, yield an excellent agreement with the measured profile after some adjustment of the initial mean effective backscattering decay lengths and, sometimes, after a slight change of the backscattering factors given by the Ichimura–Shimizu relations. The backscattering effect is studied as a function of the single layer thickness. A critical layer thickness can be determined, below which the backscattering influence becomes negligible for typical AES depth profiling results. Copyright © 2007 John Wiley & Sons, Ltd.     

Hydrophilic interaction ultra performance liquid chromatography retention prediction under gradient elution

The development and application of new separation mechanisms such as hydrophilic interaction chromatography (HILIC) is of high importance for the simultaneous analysis of polar molecules such as primary metabolites. However the retention mechanism in HILIC is not fully understood and as a result retention prediction tools are not at hand for this chromatographic approach. In the present report we study the utility of a simple algorithm, based on a simple linear and/or a simple logarithmic retention model, for retention prediction in HILIC gradient separation of a mixture of 23 selected compounds including (poly)amines, amino acids, saccharides, and other molecules. Utilizing two types of gradient elution programs with or without an isocratic part, retention data were collected in order to build prediction models. Starting from at least three gradient runs the prediction of analyte retention was very satisfactory for all gradient programs tested, providing useful evidence of the value of such retention time prediction methodologies.