A Method to Predict Blood-Brain Barrier Permeability of Drug-Like Compounds Using Molecular Dynamics Simulations

The blood-brain barrier (BBB) is formed by specialized tight junctions between endothelial cells that line brain capillaries to create a highly selective barrier between the brain and the rest of the body. A major problem to overcome in drug design is the ability of the compound in question to cross the BBB. Neuroactive drugs are required to cross the BBB to function. Conversely, drugs that target other parts of the body ideally should not cross the BBB to avoid possible psychotropic side effects. Thus, the task of predicting the BBB permeability of new compounds is of great importance. Two gold-standard experimental measures of BBB permeability are logBB (the concentration of drug in the brain divided by concentration in the blood) and logPS (permeability surface-area product). Both methods are time-consuming and expensive, and although logPS is considered the more informative measure, it is lower throughput and more resource intensive. With continual increases in computer power and improvements in molecular simulations, in silico methods may provide viable alternatives. Computational predictions of these two parameters for a sample of 12 small molecule compounds were performed. The potential of mean force for each compound through a 1,2-dioleoyl-sn-glycero-3-phosphocholine bilayer is determined by molecular dynamics simulations. This system setup is often used as a simple BBB mimetic. Additionally, one-dimensional position-dependent diffusion coefficients are calculated from the molecular dynamics trajectories. The diffusion coefficient is combined with the free energy landscape to calculate the effective permeability (P_eff) for each sample compound. The relative values of these permeabilities are compared to experimentally determined logBB and logPS values. Our computational predictions correlate remarkably well with both logBB (R² = 0.94) and logPS (R² = 0.90). Thus, we have demonstrated that this approach may have the potential to provide reliable, quantitatively predictive BBB permeability, using a relatively quick, inexpensive method.     

A Method to Predict Blood-Brain Barrier Permeability of Drug-Like Compounds Using Molecular Dynamics Simulations

The blood-brain barrier (BBB) is formed by specialized tight junctions between endothelial cells that line brain capillaries to create a highly selective barrier between the brain and the rest of the body. A major problem to overcome in drug design is the ability of the compound in question to cross the BBB. Neuroactive drugs are required to cross the BBB to function. Conversely, drugs that target other parts of the body ideally should not cross the BBB to avoid possible psychotropic side effects. Thus, the task of predicting the BBB permeability of new compounds is of great importance. Two gold-standard experimental measures of BBB permeability are logBB (the concentration of drug in the brain divided by concentration in the blood) and logPS (permeability surface-area product). Both methods are time-consuming and expensive, and although logPS is considered the more informative measure, it is lower throughput and more resource intensive. With continual increases in computer power and improvements in molecular simulations, in silico methods may provide viable alternatives. Computational predictions of these two parameters for a sample of 12 small molecule compounds were performed. The potential of mean force for each compound through a 1,2-dioleoyl-sn-glycero-3-phosphocholine bilayer is determined by molecular dynamics simulations. This system setup is often used as a simple BBB mimetic. Additionally, one-dimensional position-dependent diffusion coefficients are calculated from the molecular dynamics trajectories. The diffusion coefficient is combined with the free energy landscape to calculate the effective permeability (P_eff) for each sample compound. The relative values of these permeabilities are compared to experimentally determined logBB and logPS values. Our computational predictions correlate remarkably well with both logBB (R² = 0.94) and logPS (R² = 0.90). Thus, we have demonstrated that this approach may have the potential to provide reliable, quantitatively predictive BBB permeability, using a relatively quick, inexpensive method.     

Using Molecular Dynamics to Predict Factors Affecting Binding Strength and Magnetic Relaxivity of MRI Contrast Agents

We demonstrate the use of molecular dynamics and molecular mechanics methods to calculate properties and behavior of metal-chelate complexes that can be used as MRI contrast agents. Static and dynamic properties of several known agents were calculated and compared with experiment. We calculated the static properties such as the q-values (number of inner shell waters) and binding distances of chelate atoms to the metal ion for a set of chelates with known X-ray structure. The dynamic flexibility of the chelate arms was also calculated. These computations were extended to a series of exploratory chelate structures in order to estimate their potential as MRI contrast agents. We have also calculated for the first time the NMR relaxivity of an MRI contrast agent using a long (5 nsec) molecular dynamics simulation. Our predictions are promising enough that the method should prove useful for evaluating novel candidate compounds before they are synthesized. One novel static property, the projected area of chelate atoms onto a virtual surface centered on the metal ion (gnomonic projection), was found to give an effective measure of how well the chelate atoms use the free space around the metal ion.     

Modelling Metal Complexes of Calixarene Esters and Phosphine Oxides Using Molecular Mechanics

This paper focuses on the molecular modelling of a number of calixarene ester and phosphine oxide metal ion complexes. Monte Carlo conformational searches, in conjunction with the Merck Molecular Force Field, were carried out using Spartan SGI Version 5.0.1. running on Silicon Graphics O2 workstations. In the case of the calix[4]arene tetraesters, the optimised models strongly suggest that the selectivity of these ligands is strongly related to the eight-fold nature of the coordination with the Na⁺ ion, while coordination with the Li⁺ ion, for example, is merely three-fold. This feature of eight-fold coordination is also observed in the models of the complexes formed by the calix[4]arene tetraphosphine oxides with calcium. However, whereas the eight-fold coordination is unique to the model of the TPOL:Ca²⁺ complex among the ions modelled, this mode of coordination occurs for TPOS with sodium and potassium, in addition to calcium. This concurs with the observation that calcium selectivity is obtained with ion selective electrodes based on TPOL but not TPOS. Though the cavity in the calix[5]arenes PPOL and PPOLx and the calix[6]arene HPOL, in their uncomplexed form, are much larger than that of the corresponding calix[4]arenes, the pattern of selectivity is the same – the ligands are selective for calcium. The models of the complexes of these larger calixarenes, such as PPOL:Ca²⁺, strongly suggest that the reason for this similarity is that four of the available phosphine oxide groups complex with the calcium ion, and the others are forced away from the cavity region for steric reasons. The resulting eight-fold coordination, is therefore, similar to that of the calix[4]arenes studied.Electronic Supplementary Material available.     

Using Over-Represented Tetrapeptides to Predict Protein Submitochondria Locations

The mitochondrion is a key organelle of eukaryotic cell that provides the energy for cellular activities. Correctly identifying submitochondria locations of proteins can provide plentiful information for understanding their functions. However, using web-experimental methods to recognize submitochondria locations of proteins are time-consuming and costly. Thus, it is highly desired to develop a bioinformatics method to predict the submitochondria locations of mitochondrion proteins. In this work, a novel method based on support vector machine was developed to predict the submitochondria locations of mitochondrion proteins by using over-represented tetrapeptides selected by using binomial distribution. A reliable and rigorous benchmark dataset including 495 mitochondrion proteins with sequence identity ≤25 % was constructed for testing and evaluating the proposed model. Jackknife cross-validated results showed that the 91.1 % of the 495 mitochondrion proteins can be correctly predicted. Subsequently, our model was estimated by three existing benchmark datasets. The overall accuracies are 94.0, 94.7 and 93.4 %, respectively, suggesting that the proposed model is potentially useful in the realm of mitochondrion proteome research. Based on this model, we built a predictor called TetraMito which is freely available at http://lin.uestc.edu.cn/server/TetraMito.     

Translating Cancer Molecular Variability into Personalized Information Using Bulk and Single Cell Approaches

Cancer research is striving toward new frontiers of assigning the correct personalized drug(s) to a given patient. However, extensive tumor heterogeneity poses a major obstacle. Tumors of the same type often respond differently to therapy, due to patient‐specific molecular aberrations and/or untargeted tumor subpopulations. It is frequently not possible to determine a priori which patients will respond to a certain therapy or how an efficient patient‐specific combined therapy should be designed. Large‐scale datasets have been growing at an accelerated pace and various technologies and analytical tools for single cell and bulk level analyses are being developed to extract significant individualized signals from such heterogeneous data. However, personalized therapies that dramatically alter the course of the disease remain scarce, and most tumors still respond poorly to medical care. In this review, the basic concepts of bulk and single cell approaches are discussed, as well as their emerging role in individualized designs of drug therapies, including the advantages and limitations of their applications in personalized medicine.     

Assessment of Stalk Properties to Predict Juice Yield in Sorghum

Sweet sorghum is an outstanding feedstock choice for bioethanol production, but the gap between theoretical and commercial ethanol yields must be reduced to improve economic viability. Extractable juice yield is a primary limiting factor for higher ethanol yield, but current phenotyping techniques to measure juice yield in sorghum can be laborious. Therefore, alternative approaches to measuring juice yield during selection are needed. The objectives of this study were to investigate the relationship between stalk-related traits and juice yield and to assess the ability to predict juice yield using agronomic traits and stalk properties across and within a diverse set of sorghum ideotypes (photoinsensitive, photosensitive, biomass, grain, and sweet types). Stalk weight, stalk volume, stalk diameter, and plant height had significantly strong associations with juice yield, which were consistent across different sorghum ideotypes. The direct and indirect effects of multiple predictive traits on juice yield varied greatly with the distinct sorghum subsets. However, equation modeling demonstrated that juice yield is satisfactorily predicted by jointly assessing stalk weight and stalk moisture. Moreover, alternative prediction models involving distinct combinations of agronomic and stalk-related traits had similarly good prediction accuracy. Altogether, this suggests that several prediction models can be used to accelerate phenotyping for juice yield, which will improve the selection process. Overall, the results indicate that increasing sorghum juice yield via indirect selection is possible, but the choice of prediction model depends on the ideotypes and resources available in a breeding program.     

Using Highly Detailed Administrative Data to Predict Pneumonia Mortality

Background

Mortality prediction models generally require clinical data or are derived from information coded at discharge, limiting adjustment for presenting severity of illness in observational studies using administrative data.

Objectives

To develop and validate a mortality prediction model using administrative data available in the first 2 hospital days.

Research Design

After dividing the dataset into derivation and validation sets, we created a hierarchical generalized linear mortality model that included patient demographics, comorbidities, medications, therapies, and diagnostic tests administered in the first 2 hospital days. We then applied the model to the validation set.

Subjects

Patients aged ≥18 years admitted with pneumonia between July 2007 and June 2010 to 347 hospitals in Premier, Inc.’s Perspective database.

Measures

In hospital mortality.

Results

The derivation cohort included 200,870 patients and the validation cohort had 50,037. Mortality was 7.2%. In the multivariable model, 3 demographic factors, 25 comorbidities, 41 medications, 7 diagnostic tests, and 9 treatments were associated with mortality. Factors that were most strongly associated with mortality included receipt of vasopressors, non-invasive ventilation, and bicarbonate. The model had a c-statistic of 0.85 in both cohorts. In the validation cohort, deciles of predicted risk ranged from 0.3% to 34.3% with observed risk over the same deciles from 0.1% to 33.7%.

Conclusions

A mortality model based on detailed administrative data available in the first 2 hospital days had good discrimination and calibration. The model compares favorably to clinically based prediction models and may be useful in observational studies when clinical data are not available.     

A Computer-Aided Radiopharmaceutical Drug Design Study Using Ab Initio and Molecular Mechanics Methods

This is an investigation of technetium ligands and their complexes with [TcO]³⁺ using ab initio population analysis and molecular mechanics conformational searching methods. Calculated atomic electronic populations on the technetium atom in complexes with a number of ligands gauge the degree of covalent bonding between technetium and these ligands. Here a reduction in the positive charge on the [TcO]³⁺ moiety by complexation with a given ligand is correlated with covalent bonding. Our ab initio results suggest that ligands with more sulphur atoms have better covalent bonding to technetium than do other ligands. A conformational analysis of the uncomplexed ligands indicates that conformational reorganization before complexation correlates inversely with stable complex formation. This conformational analysis shows that ligands with ethylene carbonyl bridges have low energy conformations closer to the final complexation geometries than do ligands with ethylene, propylene or propylene carbonyl bridges. The presence of these low energy conformations facilitates a faster complexation of the ethylene carbonyl [TcO]³⁺ moiety. This result produces a kinetic explaination why ethylene carbonyl bridged ligands form stable complexes while many other ligands do not [1]. The conclusion is that kinetic and thermodynamic considerations play a role in stable complex formation between these ligands and technetium.     

WallGen,Software to Construct Layered Cellulose-Hemicellulose Networks and Predict Their Small Deformation Mechanics

We understand few details about how the arrangement and interactions of cell wall polymers produce the mechanical properties of primary cell walls. Consequently, we cannot quantitatively assess if proposed wall structures are mechanically reasonable or assess the effectiveness of proposed mechanisms to change mechanical properties. As a step to remedying this, we developed WallGen, a Fortran program (available on request) building virtual cellulose-hemicellulose networks by stochastic self-assembly whose mechanical properties can be predicted by finite element analysis. The thousands of mechanical elements in the virtual wall are intended to have one-to-one spatial and mechanical correspondence with their real wall counterparts of cellulose microfibrils and hemicellulose chains. User-defined inputs set the properties of the two polymer types (elastic moduli, dimensions of microfibrils and hemicellulose chains, hemicellulose molecular weight) and their population properties (microfibril alignment and volume fraction, polymer weight percentages in the network). This allows exploration of the mechanical consequences of variations in nanostructure that might occur in vivo and provides estimates of how uncertainties regarding certain inputs will affect WallGen''s mechanical predictions. We summarize WallGen''s operation and the choice of values for user-defined inputs and show that predicted values for the elastic moduli of multinet walls subject to small displacements overlap measured values. “Design of experiment” methods provide systematic exploration of how changed input values affect mechanical properties and suggest that changing microfibril orientation and/or the number of hemicellulose cross-bridges could change wall mechanical anisotropy.Plant scientists have long studied how primary wall structure influences mechanical properties (Preston, 1974). In this work, we develop methods to predict the elastic modulus for layered networks of cellulose microfibrils (CMFs) cross-linked by hemicellulose (HC) chains when they are subject to small imposed displacements.Polysaccharides provide over 90% of wall mass and therefore are likely to dominate wall mechanics. Two distinct but probably interacting (Zykwinska et al., 2005) networks are recognized: a cellulose-hemicellulose (CHC) network and a pectin network. Pectins can be removed by mutations, allowing measurements of the mechanical properties of the CHC network (Ryden et al., 2003) that can be compared with predicted values. The two networks probably make roughly comparable mechanical contributions in pectin-rich dicots (Ryden et al., 2003), but the CHC network presumably dominates in monocots with pectin-poor, type II walls (Carpita and Gibeaut, 1993; Rose, 2003). Plant cells align CMFs (Baskin, 2005) but not noncellulosic polysaccharides such as pectins and HCs. CMF alignment, therefore, underlies the structural and mechanical anisotropy seen in many cell walls.In principle, wall structure can predict mechanical properties, a multiscale modeling problem of the type that materials scientists often tackle (Kwon et al., 2008). In this context, structural and mechanical inputs concern polymer chains or aggregates, and mechanical properties are predicted for pieces of material several orders of magnitude larger that contain many polymer chains. There are some structure-based quantitative predictions of the mechanics of secondary walls (Bergander and Salmén, 2002; Keckes et al., 2003; Salmén, 2004; Hofstetter et al., 2005; Altaner and Jarvis, 2008), but most discussions of primary walls only involve qualitative consideration of how factors such as CMF length and alignment might change growth anisotropy (Wasteneys, 2004; Baskin, 2005) rather than the small displacement mechanical properties with which we are concerned. Modeling plant cell walls provides several particular challenges. First, walls vary greatly in CMF alignment, with multinet, polylamellate, helicoidal, and other types recognized; second, polymer composition varies even within one wall type; and third, polymer interactions remain uncertain, with the view that HCs cross-bridge CMFs (Hayashi, 1989) challenged on various grounds by those regarding them as providing spacing or otherwise facilitating movement between CMFs (Whitney et al., 1999; Thompson, 2005). In beginning multiscale modeling of primary walls, therefore, we sought a strategy that facilitated in silico experiments in which we could vary the structure, composition, and other wall properties that contribute to the complex microstructure of cell walls and that provided the opportunity to give the polymers more complex properties in future studies.Many modeling strategies facilitate computation by simplifying (homogenizing) wall structure by aggregating the properties of many polymers. Procedures are well established but do not obviate the necessity to understand the underlying polymer properties and impose an additional requirement to deduce the properties of the population being homogenized. Cell walls are often compared with fiber composites, for which several approaches to the prediction of the elastic properties have been reported (Chamis and Sendeckyj, 1968). Most such micromechanical approaches, however, are based on simplifying assumptions about the geometry of the microstructure or special relations between the phase properties. Moreover, although cell walls are often described as fiber composites, this obscures important distinctions, notably the difference between the continuous interfiber matrix typical of most manufactured fiber composites and the discrete HC cross-links present in the cell wall. The mechanical properties of the continuous matrix are relatively easily measured for manufactured fiber composites, but replacing HC cross-bridges with a continuous matrix requires defining its mechanical properties. Various micromechanical models of secondary walls assume that a homogenous HC matrix surrounds CMFs (Bergander and Salmén, 2002; Salmén, 2004; Hofstetter et al., 2005); for example, Hofstetter et al. (2005) gave this phase a bulk modulus taken from testing an isotropic HC powder. Increased computing power now provides the option to avoid such homogenization with at least three advantages accruing. First, homogenization often limits the ease with which different structures can be investigated (a high priority issue for us), since homogenization assumptions may need to be reexamined and recalibrated as microstructure changes. Without homogenization, a wide range of structures can be analyzed, given that a flexible system is available for generating microstructure. Second, accumulating knowledge of the mechanics of individual polymer chains coming from techniques such as atomic force microscopy can be directly applied to the individual HCs and CMFs in a nonhomogenized model. If homogenization is applied, that relationship is lost and new assumptions must be made about the properties of the population. Third, once the basic model is established, the properties of the polymers, particularly those of the HCs, can be varied to more accurately capture the nonlinear and other properties seen on extension.We avoided homogenization by using the WallGen program to build a fragment of virtual wall whose components have one-to-one spatial and mechanical correspondence with the CMFs and HCs of a primary wall CHC network. We chose finite element analysis (FEA) to predict the mechanical properties of the entire fragment containing thousands of CMFs and HCs. In effect, then, WallGen averages by setting up the most realistic spatial arrangement, using mechanical data for individual chains and leaving FEA to predict the collective properties. The well-established engineering technique of FEA has been used to predict wall mechanics at cellular and subcellular scales. Examples include predicting cell response to microindentation (Bolduc et al., 2006) or compression between flat plates (Smith et al., 1998) and predicting the mechanics of pulped fiber networks in paper (Hansson and Rasmuson, 2004). These applications have not involved mechanical representation of individual wall polymers, but FEA has been used at this scale to model individual microtubules and F-actin polymers pulling on membranes (Allen et al., 2009) and at even finer scales to model tubulin lattice deformation within single microtubules (Schaap et al., 2006). Modern FEA programs have features of potential value for developing more sophisticated models of wall mechanics: components can have nonlinear force-extension properties and viscoelastic properties, and conditions can be specified to break links between components of the microstructure. This should allow exploration of the more complex mechanical behavior that CHC networks show when subject to larger displacements and incorporation of additional mechanical elements providing the properties generated by pectins.In this article, we describe how WallGen operates, review the choice of values for several important inputs, predict the elastic moduli of multinet walls in which HCs cross-link CMFs, compare those values with experimental values, and quantify the mechanical effects of varying several inputs to the virtual wall. We restrict consideration to polymers given linear elastic properties and, because small strains are sufficient to predict the elastic modulus, restrict experiments to small displacements to minimize inaccuracies from this simplification. A previous publication considered issues relating to representative volume elements and analyzed some simpler CHC networks (Kha et al., 2008).     

Using Optics to Measure Biological Forces and Mechanics

Spanning all size levels, regulating biological forces and transport are fundamental life processes. Used by various investigators over the last dozen years, optical techniques offer unique advantages for studying biological forces. The most mature of these techniques, optical tweezers, or the single-beam optical trap, is commercially available and is used by numerous investigators. Although technical innovations have improved the versatility of optical tweezers, simple optical tweezers continue to provide insights into cell biology. Two new, promising optical technologies, laser-tracking microrheology and the optical stretcher, allow mechanical measurements that are not possible with optical tweezers. Here, I review these various optical technologies and their roles in understanding mechanical forces in cell biology.     

Using Genome-Wide Protein Sequence Data to Predict Amino Acid Conservation

For most proteins, multiple sequence alignments are a viable method to identify functionally and structurally important amino acids, but for most organisms, there is a subset of proteins that are unique or found in a few closely related organisms. For these proteins, it is not possible to produce sequence alignments that are useful in identifying functionally or structurally important amino acids. We have investigated the relationship between amino acid conservation and five factors (the amino acid’s identity, N-terminal neighbor, C-terminal neighbor, the local hydropathy of surrounding amino acids, and the local expected net charge of the surrounding amino acids based on the primary sequence) in Escherichia coli proteins. For four of the factors examined (all but the amino acid’s identity), there is a significant relationship with conservation for some of the standard 20 amino acids. Using the combination of all five factors, we show that it is possible to calculate a score based on the primary sequences of a subset of E. coli proteins that has statistically significant predictive value with respect to predicting conserved amino acids in other E. coli proteins and Saccharomyces cerevisiae proteins. As these five variables show significant relationships with conservation, we have termed them conservation factors. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

The Molecular Mechanics of Quantized Valence Bonds

A new force field, Quantized Valence Bonds′ Molecular Mechanics (QVBMM) has been included in the molecular modeling program STR3DI.EXE. The QVBMM force field successfully embraces and implements all of the pivotal concepts in VSEPR theory and uniquely integrates lone pairs into molecular mechanics. QVBMM facilitates a detailed analysis of the stereo-electronic effects that contribute to the structural and conformational preferences of organic molecules in their ground states, including those molecules that possess the common heteroatoms. The design, parameterization and application of the force field to a few representative molecules is discussed. The anomeric effect is also briefly examined.     

Using Data-Driven Rules to Predict Mortality in Severe Community Acquired Pneumonia

Prediction of patient-centered outcomes in hospitals is useful for performance benchmarking, resource allocation, and guidance regarding active treatment and withdrawal of care. Yet, their use by clinicians is limited by the complexity of available tools and amount of data required. We propose to use Disjunctive Normal Forms as a novel approach to predict hospital and 90-day mortality from instance-based patient data, comprising demographic, genetic, and physiologic information in a large cohort of patients admitted with severe community acquired pneumonia. We develop two algorithms to efficiently learn Disjunctive Normal Forms, which yield easy-to-interpret rules that explicitly map data to the outcome of interest. Disjunctive Normal Forms achieve higher prediction performance quality compared to a set of state-of-the-art machine learning models, and unveils insights unavailable with standard methods. Disjunctive Normal Forms constitute an intuitive set of prediction rules that could be easily implemented to predict outcomes and guide criteria-based clinical decision making and clinical trial execution, and thus of greater practical usefulness than currently available prediction tools. The Java implementation of the tool JavaDNF will be publicly available.     

Using Bacteria to Determine Protein Kinase Specificity and Predict Target Substrates

The identification of protein kinase targets remains a significant bottleneck for our understanding of signal transduction in normal and diseased cellular states. Kinases recognize their substrates in part through sequence motifs on substrate proteins, which, to date, have most effectively been elucidated using combinatorial peptide library approaches. Here, we present and demonstrate the ProPeL method for easy and accurate discovery of kinase specificity motifs through the use of native bacterial proteomes that serve as in vivo libraries for thousands of simultaneous phosphorylation reactions. Using recombinant kinases expressed in E. coli followed by mass spectrometry, the approach accurately recapitulated the well-established motif preferences of human basophilic (Protein Kinase A) and acidophilic (Casein Kinase II) kinases. These motifs, derived for PKA and CK II using only bacterial sequence data, were then further validated by utilizing them in conjunction with the scan-x software program to computationally predict known human phosphorylation sites with high confidence.