An Artificial Intelligence-Supported Medicinal Chemistry Project: An Example for Incorporating Artificial Intelligence Within the Pharmacy Curriculum.

OBJECTIVE: This study aims to integrate and use AI to teach core concepts in a medicinal chemistry course and to increase the familiarity of pharmacy students with AI in pharmacy practice and drug development. Artificial intelligence (AI) is a multidisciplinary science that aims to build software tools that mimic human intelligence. AI is revolutionizing pharmaceutical research and patient care. Hence, it is important to include AI in pharmacy education to prepare a competent workforce of pharmacists with skills in this area. METHODS: AI principles were introduced in a required medicinal chemistry course for first-year pharmacy students. An AI software, KNIME, was used to examine structure-activity relationships for 5 drugs. Students completed a data sheet that required comprehension of molecular structures and drug-protein interactions. These data were then used to make predictions for molecules with novel substituents using AI. The familiarity of students with AI was surveyed before and after this activity. RESULTS: There was an increase in the number of students indicating familiarity with use of AI in pharmacy (before vs after: 25.3% vs 74.5%). The introduction of AI stimulated interest in the course content (> 60% of students indicated increased interest in medicinal chemistry) without compromising the learning outcomes. Almost 70% of students agreed that more AI should be taught in the PharmD curriculum. CONCLUSION: This is a successful and transferable example of integrating AI in pharmacy education without changing the main learning objectives of a course. This approach is likely to stimulate student interest in AI applications in pharmacy.

Pharmacokinetic model of human exposure to ciprofloxacin through consumption of fish.

Fluoroquinolones are broad-spectrum antibiotics that accumulate in the environment. To assess human exposure through the food chain, we developed a pharmacokinetic model of fluoroquinolone accumulation in fish and a human pharmacokinetic model to predict gastrointestinal concentrations of ciprofloxacin, a common fluoroquinolone, following consumption of fish. At 70 ng/L ciprofloxacin, the average in North American surface waters, the fish steady-state concentration was calculated to be 7.5 × 10-6 µg/g. Upon human consumption of the FDA-recommended portion of 113 g of fish containing this ciprofloxacin level, the predicted human intestinal concentration was 2 × 10-6 µg/mL. At 4 × 106 ng/L (4 µg/mL) ciprofloxacin, the highest recorded environmental measurement, these numbers were 0.42 µg/g in fish and 0.1 µg/mL in the human intestine. Thus, based on the ciprofloxacin MIC for E. coli of 0.13 µg/mL, background environmental ciprofloxacin levels are unlikely to be problematic, but environmental pollution can result in high intestinal levels that may cause gut dysbiosis and antibiotic resistance.

Teaching of drug disposition using physiologically based pharmacokinetic modeling software: GastroPlus as an educational tool.

Physiologically based pharmacokinetic (PBPK) modeling requires an understanding of chemical, physiologic, and pharmacokinetic principles. Active learning with PBPK modeling software (GastroPlus) may be useful to teach these scientific principles while also teaching software operation. To examine this issue, a graduate-level course was designed using learning objectives in science, software use, and PBPK model application. These objectives were taught through hands-on PBPK modeling to answer clinically relevant questions. Students demonstrated proficient use of software, based on their responses to these questions, and showed an improved understanding of scientific principles on a pre- and post-course assessment. These outcomes support the effectiveness of simultaneous teaching of interdependent software and science.NEW & NOTEWORTHY Physiologically based pharmacokinetic (PBPK) modeling is a major growth area in drug development, regulatory submissions, and clinical applications. There is a demand for experts in this area with multidisciplinary backgrounds. In this article, we describe a course designed to teach PBPK modeling and the underlying scientific principles in parallel.

Prediction of Peptide and TCR CDR3 Loops in Formation of Class I MHC-Peptide-TCR Complexes Using Molecular Models with Solvation.

Formation of major histocompatibility (MHC)-peptide-T cell receptor (TCR) complexes is central to initiation of an adaptive immune response. These complexes form through initial stabilization of the MHC fold via binding of a short peptide, and subsequent interaction of the TCR to form a ternary complex, with contacts made predominantly through the complementarity-determining region (CDR) loops of the TCR. Stimulation of an immune response is central to cancer immunotherapy. This approach depends on identification of the appropriate combinations of MHC molecules, peptides, and TCRs to elicit an antitumor immune response. This prediction is a current challenge in computational biochemistry. In this chapter, we introduce a predictive method that involves generation of multiple peptides and TCR CDR 3 loop conformations, solvation of these conformers in the context of the MHC-peptide-TCR ternary complex, extraction of parameters from the generated complexes, and use of an AI model to evaluate the potential for the assembled ternary complex to support an immune response.

Assessment of the impact of co-curricular activities on achievement of Doctor of Pharmacy program outcomes.

INTRODUCTION: Co-curricular activities are recognized as an increasingly important aspect of pharmacy education. However, the impact of these activities on student learning is not well understood compared to that of curricular learning. The purpose of this study was to assess student-perceived progress in achieving program outcomes through voluntary co-curricular activities compared with learning of the same outcomes through mandatory curricular activities. METHODS: The study was performed over six semesters between fall 2017 and spring 2020 at the University of Southern California School of Pharmacy. Separate surveys were sent to all first- through third-year doctor of pharmacy students each semester to assess the impact of curricular and co-curricular activities on improvement in six program outcomes. Graduating student survey data were also mapped to learning outcomes to assess improvement of these outcomes upon graduation. RESULTS: Three main results emerged from these data. First, there was greater variation in the impact of co-curricular activities on different learning outcomes compared to the effect of curricular activities on the same outcomes. Second, co-curricular activities had a greater impact on "soft skills," including leadership and professionalism, compared to concrete knowledge in areas such as therapeutic mechanisms. Finally, the impact of co-curricular activities on most learning outcomes diminished with progression through the curriculum while the impact of curricular activities remained relatively constant. CONCLUSIONS: Student-perceived improvement in learning of program outcomes differs when based on co-curricular compared to curricular activities. These results show how these activities can complement each other in achievement of program outcomes.

Prediction of Water Distributions and Displacement at Protein-Ligand Interfaces.

The retention and displacement of water molecules during formation of ligand-protein interfaces play a major role in determining ligand binding. Understanding these effects requires a method for positioning of water molecules in the bound and unbound proteins and for defining water displacement upon ligand binding. We describe an algorithm for water placement and a calculation of ligand-driven water displacement in >9000 protein-ligand complexes. The algorithm predicts approximately 38% of experimental water positions within 1.0 Å and about 83% within 1.5 Å. We further show that the predicted water molecules can complete water networks not detected in crystallographic structures of the protein-ligand complexes. The algorithm was also applied to solvation of the corresponding unbound proteins, and this allowed calculation of water displacement upon ligand binding based on differences in the water network between the bound and unbound structures. We illustrate use of this approach through comparison of water displacement by structurally related ligands at the same binding site. This method for evaluation of water displacement upon ligand binding may be of value for prediction of the effects of ligand modification in drug design.

Integration of Clinical and Scientific Principles in the Teaching of Drug-Drug Interactions.

Evaluation of drug-drug interactions (DDIs) is an integral part of pharmacy practice worldwide. An understanding of the scientific mechanisms behind and the clinical implications of DDIs is important for proper management of pharmacotherapy. Here, we describe an integrated approach to teaching both aspects of DDIs as a standalone module in diverse course settings. These include on-campus and online delivery to international and local audiences in small and large classes. We describe the scientific, clinical, and integrated learning objectives of the module, and we show how these can be achieved through group projects based on published DDI case reports.

Lysine acetylation regulates the interaction between proteins and membranes.

Lysine acetylation regulates the function of soluble proteins in vivo, yet it remains largely unexplored whether lysine acetylation regulates membrane protein function. Here, we use bioinformatics, biophysical analysis of recombinant proteins, live-cell fluorescent imaging and genetic manipulation of Drosophila to explore lysine acetylation in peripheral membrane proteins. Analysis of 50 peripheral membrane proteins harboring BAR, PX, C2, or EHD membrane-binding domains reveals that lysine acetylation predominates in membrane-interaction regions. Acetylation and acetylation-mimicking mutations in three test proteins, amphiphysin, EHD2, and synaptotagmin1, strongly reduce membrane binding affinity, attenuate membrane remodeling in vitro and alter subcellular localization. This effect is likely due to the loss of positive charge, which weakens interactions with negatively charged membranes. In Drosophila, acetylation-mimicking mutations of amphiphysin cause severe disruption of T-tubule organization and yield a flightless phenotype. Our data provide mechanistic insights into how lysine acetylation regulates membrane protein function, potentially impacting a plethora of membrane-related processes.

Ligand-specific pharmacogenetic effects of nonsynonymous mutations.

In pharmacogenomics, variable receptor phenotypes, resulting from genetic polymorphisms, are often described as a change in protein function or regulation observed upon exposure to a drug. However, in some instances, phenotypes are defined using a class of medications rather than individual drugs. This paradigm assumes that a variation associated with a drug response phenotype will retain the magnitude and direction of the effect for other drugs with the same mechanism of action. However, nonsynonymous polymorphisms may have ligand-specific effects. The purpose of this study was to investigate the potential for point mutations to asymmetrically affect the binding of different drugs to a common target. Ligand binding data from site-directed mutagenesis studies on five G-protein coupled receptors (beta-1 and -2 adrenergic, dopamine D2, angiotensin II and mu-opioid receptor) were collected and analyzed. Binding data from 81 studies for 253 ligands with 447 mutant proteins, including 10 naturally occurring human variants, were analyzed, yielding 1989 mutation-ligand pairs. Fold change in binding affinity for mutant proteins, relative to the wild-type, for different drugs was examined for ligand-specific effects, with a fold-change difference of one or more orders of magnitude between agents considered significant. Of the mutations examined, 49% were associated with ligand-specific effects. One human variant (T164I, beta-2 adrenergic receptor) showed ligand-specific effects for antiasthmatic agents. These results indicate that ligand-specific changes in binding are a possible consequence of missense mutations. This implies that caution needs to be exercised when grouping drugs together during design or interpretation of genotype-phenotype association studies.

Experimental Validation of the ALLNOX Program for Studying Protein-Nucleic Acid Complexes.

Measurement of distances between spectroscopic labels (e.g., spin labels, fluorophores) attached to specific sites of biomolecules is an important method for studying biomolecular complexes. ALLNOX (Addition of Labels and Linkers) has been developed as a program to model interlabel distances based on an input macromolecule structure. Here, we report validation of ALLNOX using measured distances between nitroxide spin labels attached to specific sites of a protein-DNA complex. The results demonstrate that ALLNOX predicts average interspin distances that matched with values measured with pairs of labels attached at the protein and/or DNA. This establishes a solid foundation for using spin labeling in conjunction with ALLNOX to investigate complexes without high-resolution structures. With its high degree of flexibility for the label or the target biomolecule, ALLNOX provides a useful tool for investigating the structure-function relationship in a large variety of biological molecules.

A curcumin-diglutaric acid conjugated prodrug with improved water solubility and antinociceptive properties compared to curcumin.

In this work, a curcumin-diglutaric acid (CurDG) prodrug was synthesized by conjugation of curcumin with glutaric acid via an ester linkage. The water solubility, partition coefficient, release characteristics, and antinociceptive activity of CurDG were compared to those of curcumin. The aqueous solubility of CurDG (7.48 µg/mL) is significantly greater than that of curcumin (0.068 µg/mL). A study in human plasma showed that the CurDG completely releases curcumin within 2 h, suggesting the ability of CurDG to serve as a prodrug of curcumin. A hot plate test in mice showed the highest antinociceptive effect dose of curcumin at 200 mg/kg p.o., whereas CurDG showed the same effect at an effective dose of 100 mg/kg p.o., indicating that CurDG significantly enhanced the antinociceptive effect compared to curcumin. The enhanced antinociceptive effect of CurDG may be due to improved water solubility and increased oral bioavailability compared to curcumin.

Rational drug design and synthesis of new α-Santonin derivatives as potential COX-2 inhibitors.

Sesquiterpene compounds are widely known for their numerous pharmacological activities. Herein the focus of the authors was on α-Santonin, a sesquiterpene lactone from the Artemisia genus: the aim was to determine whether α-Santonin could be considered in the treatment of inflammation and pain. To this purpose, a small series of derivatives was designed and screened in silico against the enzyme COX-2 along with the parent compound. Drug-likeness parameters were also assessed. The compounds were eventually synthesized, and few were tested to determine their efficacy in the inhibition of COX-2 activity and expression. Overall, compound A2 was the only one with a detectable inhibitory potential of COX-2 activity whilst two of its ether derivatives demonstrated improved ability in the inhibition of COX-2 expression.

Physiologically Based Pharmacokinetic Modeling in Lead Optimization. 2. Rational Bioavailability Design by Global Sensitivity Analysis To Identify Properties Affecting Bioavailability.

When medicinal chemists need to improve oral bioavailability (%F) during lead optimization, they systematically modify compound properties mainly based on their own experience and general rules of thumb. However, at least a dozen properties can influence %F, and the difficulty of multiparameter optimization for such complex nonlinear processes grows combinatorially with the number of variables. Furthermore, strategies can be in conflict. For example, adding a polar or charged group will generally increase solubility but decrease permeability. Identifying the 2 or 3 properties that most influence %F for a given compound series would make %F optimization much more efficient. We previously reported an adaptation of physiologically based pharmacokinetic (PBPK) simulations to predict %F for lead series from purely computational inputs within a 2-fold average error. Here, we run thousands of such simulations to generate a comprehensive "bioavailability landscape" for each series. A key innovation was recognition that the large and variable number of p Ka's in drug molecules could be replaced by just the two straddling the isoelectric point. Another was use of the ZINC database to cull out chemically inaccessible regions of property space. A quadratic partial least squares regression (PLS) accurately fits a continuous surface to these thousands of bioavailability predictions. The PLS coefficients indicate the globally sensitive compound properties. The PLS surface also displays the %F landscape in these sensitive properties locally around compounds of particular interest. Finally, being quick to calculate, the PLS equation can be combined with models for activity and other properties for multiobjective lead optimization.

Physiologically Based Pharmacokinetic Modeling in Lead Optimization. 1. Evaluation and Adaptation of GastroPlus To Predict Bioavailability of Medchem Series.

When medicinal chemists need to improve bioavailability (%F) within a chemical series during lead optimization, they synthesize new series members with systematically modified properties mainly by following experience and general rules of thumb. More quantitative models that predict %F of proposed compounds from chemical structure alone have proven elusive. Global empirical %F quantitative structure-property (QSPR) models perform poorly, and projects have too little data to train local %F QSPR models. Mechanistic oral absorption and physiologically based pharmacokinetic (PBPK) models simulate the dissolution, absorption, systemic distribution, and clearance of a drug in preclinical species and humans. Attempts to build global PBPK models based purely on calculated inputs have not achieved the <2-fold average error needed to guide lead optimization. In this work, local GastroPlus PBPK models are instead customized for individual medchem series. The key innovation was building a local QSPR for a numerically fitted effective intrinsic clearance (CLloc). All inputs are subsequently computed from structure alone, so the models can be applied in advance of synthesis. Training CLloc on the first 15-18 rat %F measurements gave adequate predictions, with clear improvements up to about 30 measurements, and incremental improvements beyond that.

Enhancement of Curcumin Bioavailability Via the Prodrug Approach: Challenges and Prospects.

Curcumin is a natural product with many interesting pharmacological properties. However, these are offset by the particularly poor biopharmaceutical properties. The oral bioavailability of curcumin in humans is very low, mainly due to low solubility, poor stability, and extensive metabolism. This has led to multiple approaches to improve bioavailability, including administration of curcumin with metabolism inhibitors, formulation into nanoparticles, modification of the curcumin structure, and development of curcumin prodrugs. In this paper, we focus on the pharmacokinetic outcomes of these approaches. Pharmacokinetic parameters of curcumin after release from prodrugs are dependent on the linker between curcumin and the promoiety, and the release itself may depend on the physiological and enzymatic environment at the site of cleavage. This is an area in which more data are required for rational design of improved linkers. Cytotoxicity of curcumin prodrugs seems to correlate well with cellular uptake in vitro, but the in vivo relevance is uncertain. We conclude that improved experimental and theoretical models of absorption of curcumin prodrugs, development of accurate analytical methods for simultaneous measurement of plasma levels of prodrug and released curcumin, and acquisition of more pharmacokinetic data in animal models for dose prediction in humans are required to facilitate movement of curcumin prodrugs into clinical trials.

Computer Modeling of Spin Labels: NASNOX, PRONOX, and ALLNOX.

Measurement of distances between spin labels using electron paramagnetic resonance with the double electron-electron resonance (DEER) technique is an important method for evaluation of biomolecular structures. Computation of interlabel distances is of value for experimental planning, validation of known structures using DEER-measured distances, and determination of theoretical data for comparison with experiment. This requires steps of building labels at two defined sites on proteins, DNA or RNA; calculation of allowable label conformers based on rotation around torsional angles; computation of pairwise interlabel distances on a per conformer basis; and calculation of an average distance between the two label ensembles. We have described and validated two programs for this purpose: NASNOX, which permits computation of distances between R5 labels on DNA or RNA; and PRONOX, which similarly computes distances between R1 labels on proteins. However, these programs are limited to a specific single label and single target types. Therefore, we have developed a program, which we refer to as ALLNOX (Addition of Labels and Linkers), which permits addition of any label to any site on any target. The main principle in the program is to break the molecular system into a "label," a "linker," and a "target." The user can upload a "label" with any chemistry, define a "linker" based on a SMILES-like string, and then define the "target" site. The flexibility of ALLNOX facilitates theoretical evaluation of labels prior to synthesis and accommodates evaluation of experimental data in biochemical complexes containing multiple molecular types.

Structural Characterization of Membrane-Curving Proteins: Site-Directed Spin Labeling, EPR, and Computational Refinement.

Endocytosis and other membrane remodeling processes require the coordinated generation of different membrane shapes. Proteins capable of manipulating lipid bilayers mediate these events using mechanisms that are not fully understood. Progress is limited by the small number of structures solved for proteins bound to different membrane shapes and tools capable of resolving such information. However, recent studies have shown site-directed spin labeling (SDSL) in combination with electron paramagnetic resonance (EPR) to be capable of obtaining high-resolution structural information for proteins bound to different membrane shapes. This technique can be applied to proteins with no known structure or proteins with structures known in solution. By refining the data obtained by EPR with computational modeling, 3D structures or structural models of membrane-bound proteins can be generated. In this chapter, we highlight the basic considerations and steps required to investigate the structures of membrane-bound proteins using SDSL, EPR, and computational refinement.

An Integrated Spin-Labeling/Computational-Modeling Approach for Mapping Global Structures of Nucleic Acids.

The technique of site-directed spin labeling (SDSL) provides unique information on biomolecules by monitoring the behavior of a stable radical tag (i.e., spin label) using electron paramagnetic resonance (EPR) spectroscopy. In this chapter, we describe an approach in which SDSL is integrated with computational modeling to map conformations of nucleic acids. This approach builds upon a SDSL tool kit previously developed and validated, which includes three components: (i) a nucleotide-independent nitroxide probe, designated as R5, which can be efficiently attached at defined sites within arbitrary nucleic acid sequences; (ii) inter-R5 distances in the nanometer range, measured via pulsed EPR; and (iii) an efficient program, called NASNOX, that computes inter-R5 distances on given nucleic acid structures. Following a general framework of data mining, our approach uses multiple sets of measured inter-R5 distances to retrieve "correct" all-atom models from a large ensemble of models. The pool of models can be generated independently without relying on the inter-R5 distances, thus allowing a large degree of flexibility in integrating the SDSL-measured distances with a modeling approach best suited for the specific system under investigation. As such, the integrative experimental/computational approach described here represents a hybrid method for determining all-atom models based on experimentally-derived distance measurements.

The role of the C-terminal helix of U1A protein in the interaction with U1hpII RNA.

Previous kinetic investigations of the N-terminal RNA Recognition Motif (RRM) domain of spliceosomal A protein of the U1 small nuclear ribonucleoprotein particle (U1A) interacting with its RNA target U1 hairpin II (U1hpII) provided experimental evidence for a 'lure and lock' model of binding. The final step of locking has been proposed to involve conformational changes in an α-helix immediately C-terminal to the RRM domain (helix C), which occludes the RNA binding surface in the unbound protein. Helix C must shift its position to accommodate RNA binding in the RNA-protein complex. This results in a new hydrophobic core, an intraprotein hydrogen bond and a quadruple stacking interaction between U1A and U1hpII. Here, we used a surface plasmon resonance-based biosensor to gain mechanistic insight into the role of helix C in mediating the interaction with U1hpII. Truncation, removal or disruption of the helix exposes the RNA-binding surface, resulting in an increase in the association rate, while simultaneously reducing the ability of the complex to lock, reflected in a loss of complex stability. Disruption of the quadruple stacking interaction has minor kinetic effects when compared with removal of the intraprotein hydrogen bonds. These data provide new insights into the mechanism whereby sequences C-terminal to an RRM can influence RNA binding.

Electrophoretic mobility of duplex DNA cross-linked by mechlorethamine at a cytosine-cytosine mismatch pair.

The common nitrogen mustard, mechlorethamine, can form a covalent cross-link between the two bases of a cytosine-cytosine mismatch pair within a DNA duplex. The cross-linked species can be readily separated from DNA monoadducts and unreacted strands using denaturing polyacrylamide gel electrophoresis. Here, using DNA 19 mer duplexes that are mechlorethamine cross-linked at a C(4)-C(35), C(7)-C(32), C(10)-C(29), or C(13)-C(26) mismatch pair, we show that the denaturing polyacrylamide gel electrophoresis mobility of the cross-linked species is particularly sensitive to the proximity of the C-C cross-link to the duplex end. Species that are cross-linked at a C(4)-C(35) mismatch have greater mobilities than those cross-linked at C(7)-C(32) or C(13)-C(26), and the species with a central C(10)-C(29) cross-link have the lowest mobility. The mobility is also dependent on the proximity of the cross-link to a 5'-(32)P-phosphate or a 5'-fluorescein label. We interpret these results in terms of the conformational properties of the cross-linked species in the denaturing gel. The results are consistent with the retention of partial duplex character at the end proximal to the cross-link, with an influence on the mobility of the GC/AT ratio proximal to the cross-link and at the duplex end, and a small but discernible effect of the label.