Creating a Machine Learning Tool to Predict Acute Kidney Injury in African American Hospitalized Patients.

Machine learning (ML) has been used to build high-performance prediction models in the past without considering race. African Americans (AA) are vulnerable to acute kidney injury (AKI) at a higher eGFR level than Caucasians. AKI increases mortality, length of hospital stays, and incidence of chronic kidney disease (CKD) and end-stage renal disease (ESRD). We aimed to establish an ML-based prediction model for the early identification of AKI in hospitalized AA patients by utilizing patient-specific factors in an ML algorithm to create a predictor tool. This is a single-center, retrospective chart review. We included participants 18 years or older and admitted to an urban academic medical center. Two hundred participants were included in the study. Our ML training set provided a result of 77% accuracy for the prediction of AKI given the attributes collected. For the test set, AKI was accurately predicted in 71% of participants. The clinical significance of this model can lead to great advancements in the care of AA patients and provide practitioners avenues to optimize their therapy of choice in AAs when given AKI risk ahead of time.

Comparative Analysis of QSAR-based vs. Chemical Similarity Based Predictors of GPCRs Binding Affinity.

Ligand based virtual screening (LBVS) approaches could be broadly divided into those relying on chemical similarity searches and those employing Quantitative Structure-Activity Relationship (QSAR) models. We have compared the predictive power of these approaches using some datasets of compounds tested against several G-Protein Coupled Receptors (GPCRs). The k-Nearest Neighbors (kNN) QSAR models were built for known ligands of each GPCR target independently, with a fraction of tested ligands for each target set aside as a validation set. The prediction accuracies of QSAR models for making active/inactive calls for compounds in both training and validation sets were compared to those achieved by the Prediction of Activity Spectra for Substances' (PASS) and the Similarity Ensemble Approach (SEA) tools both available online. Models developed with the kNN QSAR method showed the highest predictive power for almost all tested GPCR datasets. The PASS software, which incorporates multiple end-point specific QSAR models demonstrated a moderate predictive power, while SEA, a chemical similarity based approach, had the lowest prediction power. Our studies suggest that when sufficient amount of data is available to develop and rigorously validate QSAR models such models should be chosen as the preferred virtual screening tool in ligand-based computational drug discovery as compared to chemical similarity based approaches.

Discovery of a Novel HDAC2 Inhibitor by a Scaffold-Merging Hybrid Query.

Histone deacetylases (HDACs) are part of a vast family of enzymes with crucial roles in numerous biological processes, largely through their repressive influence on transcription, with serious implications in a variety of human diseases. Among different isoforms, human HDAC2 in particular draws attention as a promising target for the treatment of cancer and memory deficits associated with neurodegenerative diseases. Now the challenge is to obtain a compound that is structurally novel and truly selective to HDAC2 because most current HDAC2 inhibitors do not show isoforms selectivity and suffer from metabolic instability. In order to identify novel, and isoform-selective inhibitors for human HDAC2, we designed a shape-based hybrid query from multiple scaffolds of known chemical classes and validated it to be a more effective approach to discover diverse scaffolds than single-molecule query. The hybrid query-based screening rendered a hit compound with the N-benzylaniline scaffold which showed moderate inhibitory activity against HDAC2, and its chemical structure is diverse compared to known HDAC2 inhibitors. Notably, this compound shows the selectivity against the HDAC6, a Class II enzyme, thus has the potential to further develop into the class- and isoform-selective inhibitors. Our present study supplies an useful approach to identifying novel HDAC2 inhibitors, and can be extended to the inquires of other important biomedical targets as well.

Recent patents on ophthalmic nanoformulations and therapeutic implications.

Nanoformulations (NF) are widely explored as potential alternatives for traditional ophthalmic formulation approaches. The effective treatment of ocular diseases using conventional eye drops is often hampered by factors such as: physiological barriers, rapid elimination, protein binding, and enzymatic drug degradation. Combined, these factors are known to contribute to reduced ocular residence time and poor bioavailability. Recent research studies demonstrated that NF can significantly enhance the therapeutic efficacy and bioavailability of ocular drugs, compared to the established ophthalmic drug delivery strategies. The research studies resulted in a number of patent inventions, reporting a significant increase in therapeutic efficacy for various chronic ocular disease states of both the anterior and posterior ocular segments. This article reviews these patent disclosures in detail and emphasizes the therapeutic advantages conferred by the following nanoformulation approaches: Calcium Phosphate (CaP) nanoparticles, Liposomes, Nanoemulsions, Nanomicelles, and Hydrogels. The nanoformulation approaches were shown to enhance the ocular bioavailability by reducing the drugprotein binding, increasing the corneal resident time, enhancing the drug permeability and providing a sustained drug release. Further, the article discusses United States Food and Drug Administration (USFDA) approved ocular drugs employing nanotechnology and future developments. It should be noted that, despite the potential therapeutic promise demonstrated by nanotechnology for ocular drug delivery, the bench to bed transition from patent inventions to marketed drug products has been insignificant. Majority of the discussed technologies are still in development and testing phase for commercial viability. Further, studies are in progress to assess ocular tolerance and nanotoxicity for prolonged use of NF.

Analysis of miniature single- and double-notch bending specimens for estimating the fracture toughness of cortical bone.

Studies of the fracture behavior of cortical bone have determined multiple toughening mechanisms that are active during propagation of a crack. Common methods for measuring bone fracture toughness use single-notched specimens often in four-point (SN4PB) or three-point bending (SN3PB). A double-notch four-point bending (DN4PB) specimen is useful to study prefailure damage at the crack tip. Total failure occurs at one notch and only partial failure at the other allowing study of prefailure damage in the unbroken notch. There is no widely known method for calculating the fracture toughness of bone using a DN4PB specimen. A method for calculating the fracture toughness of cortical bone using a DN4PB is developed here and compared with results for a common SN3PB specimen. The new double-notch method permits using a single specimen to measure apparent fracture toughness and to study both pre- and postfailure microdamage in the bone matrix. When and how to use the new and the established test specimens for understanding bone mechanics is discussed.

Cheminformatics meets molecular mechanics: a combined application of knowledge-based pose scoring and physical force field-based hit scoring functions improves the accuracy of structure-based virtual screening.

Poor performance of scoring functions is a well-known bottleneck in structure-based virtual screening (VS), which is most frequently manifested in the scoring functions' inability to discriminate between true ligands vs known nonbinders (therefore designated as binding decoys). This deficiency leads to a large number of false positive hits resulting from VS. We have hypothesized that filtering out or penalizing docking poses recognized as non-native (i.e., pose decoys) should improve the performance of VS in terms of improved identification of true binders. Using several concepts from the field of cheminformatics, we have developed a novel approach to identifying pose decoys from an ensemble of poses generated by computational docking procedures. We demonstrate that the use of target-specific pose (scoring) filter in combination with a physical force field-based scoring function (MedusaScore) leads to significant improvement of hit rates in VS studies for 12 of the 13 benchmark sets from the clustered version of the Database of Useful Decoys (DUD). This new hybrid scoring function outperforms several conventional structure-based scoring functions, including XSCORE::HMSCORE, ChemScore, PLP, and Chemgauss3, in 6 out of 13 data sets at early stage of VS (up 1% decoys of the screening database). We compare our hybrid method with several novel VS methods that were recently reported to have good performances on the same DUD data sets. We find that the retrieved ligands using our method are chemically more diverse in comparison with two ligand-based methods (FieldScreen and FLAP::LBX). We also compare our method with FLAP::RBLB, a high-performance VS method that also utilizes both the receptor and the cognate ligand structures. Interestingly, we find that the top ligands retrieved using our method are highly complementary to those retrieved using FLAP::RBLB, hinting effective directions for best VS applications. We suggest that this integrative VS approach combining cheminformatics and molecular mechanics methodologies may be applied to a broad variety of protein targets to improve the outcome of structure-based drug discovery studies.

Computational studies of ammonia channel function in glutamine 5'-phosphoribosylpyrophosphate amidotransferase.

Glutamine 5'-phosphoribosylpyrophosphate amidotransferase (GPATase) catalyzes the synthesis of 5'-phosphoribosylamine in a reaction that involves the translocation of ammonia along an intramolecular tunnel linking the two active sites of the enzyme. We now report a locally enhanced sampling (LES) strategy for modeling ammonia transfer between the active sites of Escherichia coli GPATase in its active conformation. Our calculations demonstrate that the ammonia channel in GPATase is best regarded as a "pipe" through which ammonia travels in the absence of an external "driving" potential. This combined LES/PMF computational approach, which offers a straightforward alternative to steered molecular dynamics simulations in studies of substrate channeling, also provides new insights into the molecular basis of the reduced ammonia transfer efficiency exhibited by the L415A GPATase mutant.

Discovery of geranylgeranyltransferase-I inhibitors with novel scaffolds by the means of quantitative structure-activity relationship modeling, virtual screening, and experimental validation.

Geranylgeranylation is critical to the function of several proteins including Rho, Rap1, Rac, Cdc42, and G-protein gamma subunits. Geranylgeranyltransferase type I (GGTase-I) inhibitors (GGTIs) have therapeutic potential to treat inflammation, multiple sclerosis, atherosclerosis, and many other diseases. Following our standard workflow, we have developed and rigorously validated quantitative structure-activity relationship (QSAR) models for 48 GGTIs using variable selection k nearest neighbor (kNN), automated lazy learning (ALL), and partial least squares (PLS) methods. The QSAR models were employed for virtual screening of 9.5 million commercially available chemicals, yielding 47 diverse computational hits. Seven of these compounds with novel scaffolds and high predicted GGTase-I inhibitory activities were tested in vitro, and all were found to be bona fide and selective micromolar inhibitors. Notably, these novel hits could not be identified using traditional similarity search. These data demonstrate that rigorously developed QSAR models can serve as reliable virtual screening tools, leading to the discovery of structurally novel bioactive compounds.

Novel inhibitors of human histone deacetylase (HDAC) identified by QSAR modeling of known inhibitors, virtual screening, and experimental validation.

Inhibitors of histone deacetylases (HDACIs) have emerged as a new class of drugs for the treatment of human cancers and other diseases because of their effects on cell growth, differentiation, and apoptosis. In this study we have developed several quantitative structure-activity relationship (QSAR) models for 59 chemically diverse histone deacetylase class 1 (HDAC1) inhibitors. The variable selection k nearest neighbor (kNN) and support vector machines (SVM) QSAR modeling approaches using both MolconnZ and MOE chemical descriptors generated from two-dimensional rendering of compounds as chemical graphs have been employed. We have relied on a rigorous model development workflow including the division of the data set into training, test, and external sets and extensive internal and external validation. Highly predictive QSAR models were generated with leave-one-out cross-validated (LOO-CV) q2 and external R2 values as high as 0.80 and 0.87, respectively, using the kNN/MolconnZ approach and 0.93 and 0.87, respectively, using the SVM/MolconnZ approach. All validated QSAR models were employed concurrently for virtual screening (VS) of an in-house compound collection including 9.5 million molecules compiled from the ZINC7.0 database, the World Drug Index (WDI) database, the ASINEX Synergy libraries, and other commercial databases. VS resulted in 45 structurally unique consensus hits that were considered novel putative HDAC1 inhibitors. These computational hits had several novel structural features that were not present in the original data set. Four computational hits with novel scaffolds were tested experimentally, and three of them were confirmed active against HDAC1, with IC50 values for the most active compound of 1.00 microM. The fourth compound was later identified to be a selective inhibitor of HDAC6, a Class II HDAC. Moreover, two of the confirmed hits are marketed drugs, which could potentially facilitate their further development as anticancer agents. This study illustrates the power of the combined QSAR-VS method as a general approach for the effective identification of structurally novel bioactive compounds.

Combinatorial QSAR modeling of specificity and subtype selectivity of ligands binding to serotonin receptors 5HT1E and 5HT1F.

The Quantitative Structure-Activity Relationship (QSAR) approach has been applied to model binding affinity and receptor subtype selectivity of human 5HT1E and 5HT1F receptor-ligands. The experimental data were obtained from the PDSP Ki Database. Several descriptor types and data-mining approaches have been used in the context of combinatorial QSAR modeling. Data mining approaches included k Nearest Neighbor, Automated Lazy Learning (ALL), and PLS; descriptor types included MolConnZ, MOE, DRAGON, Frequent Subgraphs (FSG), and Molecular Hologram Fingerprints (MHFs). Highly predictive QSAR models were generated for all three data sets (i.e., for ligands of both receptor subtypes and for subtype selectivity), and different individual techniques were proved best in each case. For real value activity data available for 5HT1E and 5HT1F ligand binding, models were characterized by leave-one-out cross-validated R(2) (q(2)) for the training sets and predictive R(2) values for the test sets. The best models for 5HT1E ligands were obtained with the kNN approach combined with MolConnZ descriptors (q(2)=0.69, R(2)=0.92); for 5HT1F ligands ALL QSAR method using MolConnZ descriptors gave the best results (R(2)=0.92). Rigorously validated classification models were also developed for the 5HT1E/5HT1F subtype selectivity data set with high correct classification accuracy for both training (CCRtrain=0.88) and test (CCRtest=1.00) sets using kNN with MolConnZ descriptors. The external predictive power of QSAR models was further validated by virtual screening of The Scripps Research Institute (TSRI) screening library to recover 5HT1E agonists and antagonists (not present in the original PDSP data set) with high enrichment factors. The successful development of externally predictive and interpretative QSAR models affords further design and discovery of novel subtype specific GPCR agents.

Differentiation of AmpC beta-lactamase binders vs. decoys using classification kNN QSAR modeling and application of the QSAR classifier to virtual screening.

The use of inaccurate scoring functions in docking algorithms may result in the selection of compounds with high predicted binding affinity that nevertheless are known experimentally not to bind to the target receptor. Such falsely predicted binders have been termed 'binding decoys'. We posed a question as to whether true binders and decoys could be distinguished based only on their structural chemical descriptors using approaches commonly used in ligand based drug design. We have applied the k-Nearest Neighbor (kNN) classification QSAR approach to a dataset of compounds characterized as binders or binding decoys of AmpC beta-lactamase. Models were subjected to rigorous internal and external validation as part of our standard workflow and a special QSAR modeling scheme was employed that took into account the imbalanced ratio of inhibitors to non-binders (1:4) in this dataset. 342 predictive models were obtained with correct classification rate (CCR) for both training and test sets as high as 0.90 or higher. The prediction accuracy was as high as 100% (CCR = 1.00) for the external validation set composed of 10 compounds (5 true binders and 5 decoys) selected randomly from the original dataset. For an additional external set of 50 known non-binders, we have achieved the CCR of 0.87 using very conservative model applicability domain threshold. The validated binary kNN QSAR models were further employed for mining the NCGC AmpC screening dataset (69653 compounds). The consensus prediction of 64 compounds identified as screening hits in the AmpC PubChem assay disagreed with their annotation in PubChem but was in agreement with the results of secondary assays. At the same time, 15 compounds were identified as potential binders contrary to their annotation in PubChem. Five of them were tested experimentally and showed inhibitory activities in millimolar range with the highest binding constant K(i) of 135 microM. Our studies suggest that validated QSAR models could complement structure based docking and scoring approaches in identifying promising hits by virtual screening of molecular libraries.

Stereochemical studies of the monocyclic agouti-related protein (103-122) Arg-Phe-Phe residues: conversion of a melanocortin-4 receptor antagonist into an agonist and results in the discovery of a potent and selective melanocortin-1 agonist.

The agouti-related protein (AGRP) is an endogenous antagonist of the centrally expressed melanocortin receptors. The melanocortin-4 receptor (MC4R) is involved in energy homeostasis, food intake, sexual function, and obesity. The endogenous hAGRP protein is 132 amino acids in length, possesses five disulfide bridges at the C-terminus of the molecule, and is expressed in the hypothalamus of the brain. We have previously reported that a monocyclic hAGRP(103-122) peptide is an antagonist at the melanocortin receptors expressed in the brain. Stereochemical inversion from the endogenous l- to d-isomers of single or multiple amino acid modifications in this monocyclic truncated hAGRP sequence resulted in molecules that are converted from melanocortin receptor antagonists into melanocortin receptor agonists. The Asp-Pro-Ala-Ala-Thr-Ala-Tyr-cyclo[Cys-Arg-DPhe-DPhe-Asn-Ala-Phe-Cys]-Tyr-Ala-Arg-Lys-Leu peptide resulted in a 60 nM melanocortin-1 receptor agonist that is 100-fold selective versus the mMC4R, 1000-fold selective versus the mMC3R, and ca. 180-fold selective versus the mMC5R. In attempts to identify putative ligand-receptor interactions that may be participating in the agonist induced stimulation of the MC4R, selected ligands were docked into a homology molecular model of the mMC4R. These modeling studies have putatively identified hAGRP ligand DArg111-mMC4RAsn115 (TM3) and the hAGRP DPhe113-mMC4RPhe176 (TM4) interactions as important for agonist activity.

Structural characterization and pharmacology of a potent (Cys101-Cys119, Cys110-Cys117) bicyclic agouti-related protein (AGRP) melanocortin receptor antagonist.

Agouti-related protein (AGRP) is one of two known naturally occurring antagonists of G-protein coupled receptors. AGRP is synthesized in the brain and is an antagonist of the melanocortin-3 and -4 receptors (MC3R, MC4R). These three proteins are involved in the regulation of energy homeostasis and obesity in both mice and humans. The human AGRP protein is 132 amino acids and contains five disulfide bridges in the C-terminal domain. Previous reports of the NMR structures of hAGRP(87-132) and a truncated 34 amino acid form consisting of four disulfide bridges identified that AGRP contains an inhibitor cystine knot (ICK) structural fold, and that is the first mammalian example. Herein, we report a bicyclic hAGRP analogue that, when compared to hAGRP(87-132), possesses equal binding affinity but is 80-fold less potent at the mouse MC4R. Using NMR, computer assisted molecular modeling (CAMM), and cluster analysis, we have identified five structural families, two of which are highly populated, of this bicyclic hAGRP analogue. Computational docking experiments of this bicyclic hAGRP derivative, using a three-dimensional homology molecular model of the mouse MC4R, identified that three of the five structural families could be docked into the MC4R without problems from steric hindrance. Those three docked mMC4R-bicyclic hAGRP family structures were compared with putative hAGRP(87-132) ligand-receptor interactions previously reported (Wilczynski et al. J. Med. Chem. 2004, 47, 2194) in attempts to identify a "bioactive" conformation of the bicyclic hAGRP peptide and account for the 80-fold decreased ligand potency compared to hAGRP(87-132).

Identification of putative agouti-related protein(87-132)-melanocortin-4 receptor interactions by homology molecular modeling and validation using chimeric peptide ligands.

Agouti-related protein (AGRP) is one of only two naturally known antagonists of G-protein-coupled receptors (GPCRs) identified to date. Specifically, AGRP antagonizes the brain melanocortin-3 and -4 receptors involved in energy homeostasis. Alpha-melanocyte stimulating hormone (alpha-MSH) is one of the known endogenous agonists for these melanocortin receptors. Insight into putative interactions between the antagonist AGRP amino acids with the melanocortin-4 receptor (MC4R) may be important for the design of unique ligands for the treatment of obesity related diseases and is currently lacking in the literature. A three-dimensional homology molecular model of the mouse MC4 receptor complex with the hAGRP(87-132) ligand docked into the receptor has been developed to identify putative antagonist ligand-receptor interactions. Key putative AGRP-MC4R interactions include the Arg111 of hAGRP(87-132) interacting in a negatively charged pocket located in a cavity formed by transmembrane spanning (TM) helices 1, 2, 3, and 7, capped by the acidic first extracellular loop (EL1) and specifically with the conserved melanocortin receptor residues mMC4R Glu92 (TM2), mMC4R Asp114 (TM3), and mMC4R Asp118 (TM3). Additionally, Phe112 and Phe113 of hAGRP(87-132) putatively interact with an aromatic hydrophobic pocket formed by the mMC4 receptor residues Phe176 (TM4), Phe193 (TM5), Phe253 (TM6), and Phe254 (TM6). To validate the AGRP-mMC4R model complex presented herein from a ligand perspective, we generated nine chimeric peptide ligands based on a modified antagonist template of the hAGRP(109-118) (Tyr-c[Asp-Arg-Phe-Phe-Asn-Ala-Phe-Dpr]-Tyr-NH(2)). In these chimeric ligands, the antagonist AGRP Arg-Phe-Phe residues were replaced by the melanocortin agonist His/D-Phe-Arg-Trp amino acids. These peptides resulted in agonist activity at the mouse melanocortin receptors (mMC1R and mMC3-5Rs). The most notable results include the identification of a novel subnanomolar melanocortin peptide template Tyr-c[Asp-His-DPhe-Arg-Trp-Asn-Ala-Phe-Dpr]-Tyr-NH(2) that is equipotent to alpha-MSH at the mMC1, mMC3, and mMC5 receptors but is 30-fold more potent than alpha-MSH at the mMC4R. Additionally, these studies identified a new and novel >200-fold MC4R versus MC3R selective peptide Tyr-c[Asp-D-Phe-Arg-Trp-Asn-Ala-Phe-Dpr]-Tyr-NH(2) template. Furthermore, when the His-DPhe-Arg-Trp sequence is used to replace the hAGRP Arg-Phe-Phe residues in the "mini"-AGRP (hAGRP87-120, C105A) template, a potent nanomolar agonist resulted at the mMC1R and MC3-5Rs.